Thursday, May 23, 2024

Don’t give up your day job: Generative AI and the End of Programming

Spread the love

There’s a lot of worry about software professionals “losing their jobs” to AI, to be replaced by a smarter version of ChatGPT, GitHub’s Copilot, Google’s foundation model Codey, or something similar.

Matt Welsh, the founder of an AI business, has been talking and writing about the death of programming. He’s wondering if large language models (LLMs) would destroy programming as we know it, and he’s ecstatic that the answer is “yes”: eventually, if not immediately.

But what does this actually mean? What does this entail for those who make a living by writing software?

The worth of learning new programming skills:

Some businesses will undoubtedly regard AI as a tool for replacing human effort rather than supplementing human talents. AI threatens the jobs of programmers who work for those companies. I’m sorry if you work for one of those organizations, but this is a fantastic opportunity.

Regardless of the acclaimed cutbacks, the work market for developers is perfect, it’s probably going to stay perfect, and you’re presumably in an ideal situation finding a business who doesn’t consider you to be a cost to be limited. Now is the right time to get familiar with a few new abilities and find a business who truly esteems you.

Be that as it may, the quantity of developers who are “supplanted by simulated intelligence” will be little. Here’s the reason, and this is the way the utilization of computer based intelligence will change the discipline overall. I did an exceptionally non-logical investigation of how much time software engineers really spend composing code.

Alright, I recently composed “The amount of a product designer’s time is spent coding” into the hunt bar and took a gander at the main few articles, which gave rates going from 10% to 40%. My own sense, from conversing with and noticing many individuals throughout the long term, falls into the lower end of that reach: 15% to 20%.

Time for “the reminder of the job”

What does that mean in terms of time and effort? Although I’ve seen numbers as high as 80%, I don’t think they’re accurate; instead, I think 25% to 50% is a more reasonable range. If you spend 20% of your time coding and AI-based code creation increases your productivity by 50%, then you really only gain back approximately 10% of your time.

You can utilize it to generate more code; I’ve yet to come across a coder who wasn’t overworked or working under insane deadline pressure. Alternately, you may devote more time to the “rest of the job,” which accounts for 80% of your time and wasn’t spent writing code.

While some of that time is wasted in fruitless meetings, the majority of the rest of the job is spent figuring out what the user actually needs something they didn’t tell you the first time understanding their needs designing testing debugging reviewing code, and so forth. The list is really long.

Need for programmers: AI requires skill set and design experties

Our industry has never been very adept at the “rest of the job” (especially the “user’s needs” aspect). Designing software, user interfaces, and data representations is undoubtedly here to stay and is not something the current generation of AI is very adept at.

Despite how far we’ve gone, I don’t think anyone has ever had to save code that was more accurately described as a “seething mass of bits.” If you’ve used ChatGPT frequently, you probably already know that testing and debugging won’t go away. AIs produce wrong code, and this is not going to change anytime soon.

Since it’s exceedingly difficult for a programmer to comprehend the security implications of code they didn’t write, security audits will only grow more crucial, not less. The quality of the goods we offer will undoubtedly increase if we spend more time on these issues and hand off the specifics of churning out lines of code to an AI.

Requried a new type of programming

Let’s now take a very long-term perspective. Let’s assume that Welsh is correct and that programming as we know it won’t exist in 20 years, if not tomorrow. Does it actually vanish?

Tim O’Reilly saw some of my experiments with Ethan and Lilach Mollick’s AI in the classroom prompts a couple of weeks ago. In response, he said, “This prompt is really programming.” He’s correct.


Companies are hesitant to roll out new use cases despite the risks still present; instead, they are spending time on internal innovation to gain a deeper understanding of what is achievable. For instance, a recent hackathon at eBay was fully dedicated to next-generation AI.

 It was undoubtedly much more than the executive team could have possibly imagined. Every business should think about focusing on generative AI during their innovation weeks and hackathons to explore what ideas they can generate. However, we must think carefully about that.

Leave a Reply

Your email address will not be published. Required fields are marked *